Seismosoft | Earthquake Engineering Solutions

Hello,seismosoft!

I'm modelling a steel-concrete composite portal frame in X-Z plane under low cyclic loading to verify my experiment. The height and span of portal frame is 1.5m and 3m respectively.Unfortunately,the result showed that the capacity of SRC portal frame was still on the increase when it exceeded the ultimate load capacity obtaining from experiment. I found the number of elements and beam to column joints had great influence on its capacity when I modified the analysis model. I am very thanking for you if you can reply to some questions below.

1.How many elements should be subdivided at beam to column joints? As is known, the beam to column joint should usually be subdivided more elements than beam and column in some current finite element analysis programme. I have no idea about that using fiber element in my case. I hope you can give some suggestions.

2.In my analysis model, I set a great elastic module(10 times) for material to assign to the beam to column joint in order to simulate rigid joints. Is that appropriate? Because I use link element to simulate rigid joints according to help system but it has no effect.

Thank you very much!

The same answer we gave you in the past (regarding the modelling of RC beam-column joints) still holds; this is a tough subject with no easy universally agreed and applicable solution.

We do not fully understand your doubts, since you mention at the start that the number of elements influences the results (which is unfortunately expected, due to the phenomenon of localisation), but then your two queries focus on the joint modelling alone.

In any case, answering to your questions:

1) we do not believe you should be modelling your beam-column zones with a refined mesh of frame elements. If you wish to model flexibility of beam-column joints, see previous discussions on this topic in this Forum (just do a Forum search on 'joints').

2) as discussed in other Forum threads, a multiplier of at least 250 is normally employed to represent rigid elements.

Seismosoft Support

I'm glad to receive your reply and thank you very much!

I still have a question about subdivision of inelastic frames, for example, how many elements should generally be subdivided for beams and columns in a portal frame stated above?

As I known, if the element formulation is flexibility-based, the less number of elements is needed and the accuracy of result is sufficient. But the more elements must be used when the element formulation is stiffness-based. So, which theory is used for element in seismosturct?

Because I found that the reasonable hysteresis curve could be obtained when I just subdivided two elements for beam and column. However, when 5-6 elements were used for each inelastic frame, the capacity of portal frame always rise and the curve has no softening phase.

If you read the Help System (and the scientific publications it refers to) and/or carry out a search in this forum on the word "formulation", you will readily realise that the current version of SeismoStruct features stiffness-based beam elements (also known as displacement-based beam elements), and that a number of around 4-5 elements is usually required for adequate modelling.

Seismosoft Support